@tangle-network/agent-eval 0.139.3 → 0.140.0

package/CHANGELOG.md

@@ -4,6 +4,20 @@ All notable changes to `@tangle-network/agent-eval` and its sibling `agent-eval-
 
 ---
 
+## [0.140.0] - 2026-07-31 - the recursive engine runs on real providers
+
+### Changed
+
+- The DSPy bridge takes a selectable control adapter, so a model that answers in prose and fenced code rather than DSPy's `[[ ## field ## ]]` markers no longer voids a completed investigation.
+  **The bridge's `analyze` input gains a required `controlAdapter` key**: a Python `agent-eval-rpc` and its npm peer must now be the same version, and a mismatch fails with `analyze input must contain exactly [...]`.
+- A single malformed finding never voids a completed RLM case; rejections are recorded with a reason instead of raised.
+- Raise the DSPy output-token default from 4096 to 16384. 4096 is below what current coding models emit for a full findings array — glm-5.2 through an OpenAI-compatible gateway returns 8192 and the request is rejected outright, so the old default failed before any analysis ran. `maxCostUsd` remains the real spend bound.
+
+### Added
+
+- First scored run of the recursive DSPy RLM analyst on the pinned 32-case CodeTraceBench corpus: F1 0.3644 over 60 of 64 completed cases at $6.73, against the retired one-shot runner's 0.3673 at $1.21 and CodeTracer's 0.3128.
+  Recursion did not improve step localization on this task, and the artifact says so; the engine's claimed value is verified findings on an arbitrary session, which this benchmark does not measure.
+
 ## [0.139.3] - 2026-07-31 - supervisor runs under `.agent`
 
 ### Changed

package/dist/analyst/index.d.ts

@@ -1,14 +1,14 @@
 import { a as MultiLayerVerifier, l as VerifyOptions, o as Severity } from "../multi-layer-verifier-BHY1gWAc.js";
-import { A as parseFindingSubject, C as FINDING_SUBJECT_KINDS, D as FindingSubjectStringSchema, E as FindingSubjectKind, F as DefineExactCustomAnalystOptions, G as SemanticConceptJudgeOptions, I as defineCustomAnalyst, L as defineTraceAnalyst, M as DspyRlmTraceEngineOptions, N as createDspyRlmTraceEngine, O as KIND_EXPECTED_SUBJECTS, P as DefineCustomAnalystOptions, S as FINDING_SUBJECT_GRAMMAR_PROMPT, T as FindingSubject, W as SemanticConceptJudgeInput, Y as RunCritic, Z as RunTrace, _ as FindingsDiff, a as SkillUsageScanConfig, b as defaultIsMaterial, c as DEFAULT_TRACE_ANALYST_KINDS, d as IMPROVEMENT_KIND_SPEC, f as FAILURE_MODE_KIND_SPEC, g as DiffPolicy, h as emitControlIntegrityFindings, i as SkillUsageReport, j as renderFindingSubject, k as findingSubjectGrammarPromptFor, l as KNOWLEDGE_POISONING_KIND_SPEC, m as ControlIntegrityAnalyst, n as SkillUsageAnalyst, o as buildSkillUsageReport, p as CONTROL_INTEGRITY_ANALYST, r as SkillUsageRecord, s as emitSkillUsageFindings, t as SKILL_USAGE_ANALYST, u as KNOWLEDGE_GAP_KIND_SPEC, v as FindingsStore, w as FINDING_SUBJECT_SYNTAX, x as diffFindings, y as PersistedFinding } from "../skill-usage-B-BFS8M2.js";
-import { b as CustomTokenPricing, c as CostLedgerHandle } from "../cost-ledger-DKgyIWRj.js";
-import { A as ChatClient, B as SandboxSdkTransportOpts, F as CreateChatClientOpts, I as CustomTransportOpts, L as DirectProviderTransportOpts, M as ChatResponse, N as ChatTransport, P as CliBridgeTransportOpts, R as MockTransportOpts, V as createChatClient, j as ChatRequest, k as ChatCallOpts, m as JudgeInput, p as JudgeFn, z as RouterTransportOpts } from "../types-uPrS6mD-.js";
-import { _ as makeFinding, a as AnalystInputKind, c as AnalystRunInputs, d as AnalystSeverity, f as AnalystUsageReceipt, g as computeFindingId, h as ProposalFindingOrigin, i as AnalystFinding, l as AnalystRunResult, m as ProposalFinding, n as AnalystContext, o as AnalystRequirements, p as EvidenceRef, r as AnalystCost, s as AnalystRunEvent, t as Analyst, u as AnalystRunSummary, v as makeProposalFinding, x as TraceAnalysisStore } from "../types-DoEYskCd.js";
-import { a as createTraceAnalyst, c as runTraceAnalyst, d as deriveEfficiencyFindings, i as TraceAnalystDefinition, l as BehavioralAnalystOptions, n as buildDefaultAnalystRegistry, o as renderPriorFindings, r as CreateTraceAnalystOptions, s as renderUpstreamFindings, t as DefaultAnalystRegistryOptions, u as behavioralAnalyst } from "../default-registry-B8vf7Rmf.js";
-import { a as ExactAnalystRunPolicySnapshot, c as ExactAnalystSnapshot, d as ExactExecutionComponentSnapshot, i as ExactAnalystRunEvent, l as ExactCapableAnalyst, n as ExactAnalystExecutionPlanSnapshot, o as ExactAnalystRunResult, r as ExactAnalystRunCompletion, s as ExactAnalystRunSummary, t as ExactAnalystBudgetSnapshot, u as ExactExecutionComponentIdentity } from "../exact-types-MaaFcllV.js";
-import { A as ExactAnalystRunExecutionError, D as AnalystRegistryOptions, E as AnalystRegistry, M as RegistryRunOpts, N as assertExactRegistryRunOpts, O as BudgetPolicy, T as AnalystHooks, j as ExactRegistryRunOpts, k as ExactAnalystBudgetPolicy } from "../completion-verifier-IPoP4fQO.js";
-import { C as scoreAnalystFindings, S as traceStoreEvidenceResolver, _ as AnalystIssueExpectation, a as AnalystBenchmarkLabelState, b as registryBenchmarkRunner, c as AnalystBenchmarkProvenance, d as AnalystBenchmarkSummary, f as AnalystEvidenceExpectation, g as AnalystFindingScore, h as AnalystEvidenceResolver, i as AnalystBenchmarkError, l as AnalystBenchmarkResult, m as AnalystEvidenceResolutionError, n as AnalystBenchmarkDatasetRef, o as AnalystBenchmarkObservation, p as AnalystEvidenceResolution, r as AnalystBenchmarkDescriptor, s as AnalystBenchmarkOutput, t as AnalystBenchmarkCase, u as AnalystBenchmarkRunner, v as AnalystLatencyDistribution, x as runAnalystBenchmark, y as RunAnalystBenchmarkOptions } from "../benchmark-DDVdWcwA.js";
-import { r as ExternalOptimizerRunnerCommand } from "../external-optimizer-contracts-BrxY2Sli.js";
-import { a as TraceAnalysisEngineRequest, c as resolveTraceAnalystLimits, d as RawAnalystEvidence, f as RawAnalystEvidenceSchema, g as parseRawFinding, h as evidenceRefsFromRawFinding, i as TraceAnalysisEngine, l as ANALYST_SEVERITIES, m as RawAnalystFindingSchema, n as buildTraceToolsForGroup, o as TraceAnalysisEngineResult, p as RawAnalystFinding, r as DEFAULT_TRACE_ANALYST_LIMITS, s as TraceAnalystLimits, t as TraceToolGroupName, u as RAW_FINDING_SCHEMA_PROMPT } from "../tool-groups-CdYq22lX.js";
+import { A as parseFindingSubject, C as FINDING_SUBJECT_KINDS, D as FindingSubjectStringSchema, E as FindingSubjectKind, F as DefineExactCustomAnalystOptions, G as SemanticConceptJudgeOptions, I as defineCustomAnalyst, L as defineTraceAnalyst, M as DspyRlmTraceEngineOptions, N as createDspyRlmTraceEngine, O as KIND_EXPECTED_SUBJECTS, P as DefineCustomAnalystOptions, S as FINDING_SUBJECT_GRAMMAR_PROMPT, T as FindingSubject, W as SemanticConceptJudgeInput, Y as RunCritic, Z as RunTrace, _ as FindingsDiff, a as SkillUsageScanConfig, b as defaultIsMaterial, c as DEFAULT_TRACE_ANALYST_KINDS, d as IMPROVEMENT_KIND_SPEC, f as FAILURE_MODE_KIND_SPEC, g as DiffPolicy, h as emitControlIntegrityFindings, i as SkillUsageReport, j as renderFindingSubject, k as findingSubjectGrammarPromptFor, l as KNOWLEDGE_POISONING_KIND_SPEC, m as ControlIntegrityAnalyst, n as SkillUsageAnalyst, o as buildSkillUsageReport, p as CONTROL_INTEGRITY_ANALYST, r as SkillUsageRecord, s as emitSkillUsageFindings, t as SKILL_USAGE_ANALYST, u as KNOWLEDGE_GAP_KIND_SPEC, v as FindingsStore, w as FINDING_SUBJECT_SYNTAX, x as diffFindings, y as PersistedFinding } from "../skill-usage-CVVnoIx-.js";
+import { b as CustomTokenPricing, c as CostLedgerHandle } from "../cost-ledger-FuQvHxPm.js";
+import { A as ChatClient, B as SandboxSdkTransportOpts, F as CreateChatClientOpts, I as CustomTransportOpts, L as DirectProviderTransportOpts, M as ChatResponse, N as ChatTransport, P as CliBridgeTransportOpts, R as MockTransportOpts, V as createChatClient, j as ChatRequest, k as ChatCallOpts, m as JudgeInput, p as JudgeFn, z as RouterTransportOpts } from "../types-BjMFz88h.js";
+import { _ as makeFinding, a as AnalystInputKind, c as AnalystRunInputs, d as AnalystSeverity, f as AnalystUsageReceipt, g as computeFindingId, h as ProposalFindingOrigin, i as AnalystFinding, l as AnalystRunResult, m as ProposalFinding, n as AnalystContext, o as AnalystRequirements, p as EvidenceRef, r as AnalystCost, s as AnalystRunEvent, t as Analyst, u as AnalystRunSummary, v as makeProposalFinding, x as TraceAnalysisStore } from "../types-D3jh6F98.js";
+import { a as createTraceAnalyst, c as runTraceAnalyst, d as deriveEfficiencyFindings, i as TraceAnalystDefinition, l as BehavioralAnalystOptions, n as buildDefaultAnalystRegistry, o as renderPriorFindings, r as CreateTraceAnalystOptions, s as renderUpstreamFindings, t as DefaultAnalystRegistryOptions, u as behavioralAnalyst } from "../default-registry-Ci7wAAR8.js";
+import { a as ExactAnalystRunPolicySnapshot, c as ExactAnalystSnapshot, d as ExactExecutionComponentSnapshot, i as ExactAnalystRunEvent, l as ExactCapableAnalyst, n as ExactAnalystExecutionPlanSnapshot, o as ExactAnalystRunResult, r as ExactAnalystRunCompletion, s as ExactAnalystRunSummary, t as ExactAnalystBudgetSnapshot, u as ExactExecutionComponentIdentity } from "../exact-types-B0lJV3tu.js";
+import { A as ExactAnalystRunExecutionError, D as AnalystRegistryOptions, E as AnalystRegistry, M as RegistryRunOpts, N as assertExactRegistryRunOpts, O as BudgetPolicy, T as AnalystHooks, j as ExactRegistryRunOpts, k as ExactAnalystBudgetPolicy } from "../completion-verifier-D15NHYSk.js";
+import { C as scoreAnalystFindings, S as traceStoreEvidenceResolver, _ as AnalystIssueExpectation, a as AnalystBenchmarkLabelState, b as registryBenchmarkRunner, c as AnalystBenchmarkProvenance, d as AnalystBenchmarkSummary, f as AnalystEvidenceExpectation, g as AnalystFindingScore, h as AnalystEvidenceResolver, i as AnalystBenchmarkError, l as AnalystBenchmarkResult, m as AnalystEvidenceResolutionError, n as AnalystBenchmarkDatasetRef, o as AnalystBenchmarkObservation, p as AnalystEvidenceResolution, r as AnalystBenchmarkDescriptor, s as AnalystBenchmarkOutput, t as AnalystBenchmarkCase, u as AnalystBenchmarkRunner, v as AnalystLatencyDistribution, x as runAnalystBenchmark, y as RunAnalystBenchmarkOptions } from "../benchmark-DxaZfy0w.js";
+import { r as ExternalOptimizerRunnerCommand } from "../external-optimizer-contracts-nb7c_WAR.js";
+import { a as TraceAnalysisEngineRequest, c as resolveTraceAnalystLimits, d as RawAnalystEvidence, f as RawAnalystEvidenceSchema, g as parseRawFinding, h as evidenceRefsFromRawFinding, i as TraceAnalysisEngine, l as ANALYST_SEVERITIES, m as RawAnalystFindingSchema, n as buildTraceToolsForGroup, o as TraceAnalysisEngineResult, p as RawAnalystFinding, r as DEFAULT_TRACE_ANALYST_LIMITS, s as TraceAnalystLimits, t as TraceToolGroupName, u as RAW_FINDING_SCHEMA_PROMPT } from "../tool-groups-DVQTy9lq.js";
 //#region src/analyst/adapters.d.ts
 declare function liftSeverity(s: Severity): AnalystSeverity;
 interface VerifierAdapterOpts<Env> {
@@ -366,6 +366,8 @@ interface CodeTraceBlockDiagnostics {
   unresolvedBlockInteriorSteps: number[];
   /** Steps claimed by more than one block; the first block keeps the step. */
   overlappingBlockSteps: number[];
+  /** Findings dropped before expansion because their shape or evidence is invalid. */
+  rejectedFindings?: string[];
 }
 declare function emptyPublicBenchmarkRunner(): AnalystBenchmarkRunner<AnalystRunInputs>;
 declare function adaptPublicBenchmarkFindings(options: {
@@ -685,13 +687,13 @@ interface AnalystBenchmarkCommandConfig {
   resume: boolean;
 }
 declare function runAnalystBenchmarkCommand(argv: readonly string[], env?: NodeJS.ProcessEnv, dependencies?: AnalystBenchmarkCommandDependencies): Promise<number>;
-declare const ANALYST_BENCHMARK_HELP = "agent-eval analyst-benchmark\n\nRun the recursive DSPy trace analyst against public AgentRx or CodeTraceBench labels.\n\nRequired:\n  --dataset agentrx|codetracebench\n  --analyst dspy-rlm|direct        Scored analyst. Default: dspy-rlm.\n                                   'direct' is the retired one-shot runner that\n                                   produced the published evidence.\n  --labels <dataset.json|dataset.jsonl>\n  --trace-dir <one-trace-per-file OTLP JSONL directory>\n  --artifact-dir <extracted artifact root>  Required for CodeTraceBench\n  --out <new output directory>\n  --revision <full 40- or 64-character hex digest>\n  --split <dataset split>\n  --base-url <OpenAI-compatible /v1 URL>\n  --api-key-env <environment variable containing the bearer>\n  --model <provider model id>\n  --limit <positive case count>\n\nControls:\n  --resume                         Continue an interrupted run in --out\n  --seed <integer>                 Case-selection and comparison seed. Default: 0\n  --concurrency <positive integer> Parallel benchmark jobs. Default: 1\n  --repetitions <positive integer> Runs per case and runner. Default: 1\n  --max-output-tokens <positive>   Model output limit per call. Default: 4096\n  --python <executable>             Python with agent-eval-rpc[dspy]. Default: python\n  --timeout-ms <positive>          Model analyst deadline per case. Default: 300000\n  --max-cost-usd <positive>        Run-wide spend limit. Default: 5\n  --max-artifact-bytes <positive>  Final evidence bytes per case. Default: 8388608\n\nWrites result.json with every observation, metric, usage field, error, comparison,\ninput digest, artifact digest, case distribution, selected case id, and explicit\nunknown cost. Limited deterministic-hash subsets are marked non-representative.\nCompleted observations are fsynced to observations.jsonl. Shareable output is in\nresult.json and report.md. Machine-local paths, endpoint, and command are isolated\nin run.local.json.\nThe key is read from the named environment variable and is never written.";
+declare const ANALYST_BENCHMARK_HELP = "agent-eval analyst-benchmark\n\nRun the recursive DSPy trace analyst against public AgentRx or CodeTraceBench labels.\n\nRequired:\n  --dataset agentrx|codetracebench\n  --analyst dspy-rlm|direct        Scored analyst. Default: dspy-rlm.\n                                   'direct' is the retired one-shot runner that\n                                   produced the published evidence.\n  --labels <dataset.json|dataset.jsonl>\n  --trace-dir <one-trace-per-file OTLP JSONL directory>\n  --artifact-dir <extracted artifact root>  Required for CodeTraceBench\n  --out <new output directory>\n  --revision <full 40- or 64-character hex digest>\n  --split <dataset split>\n  --base-url <OpenAI-compatible /v1 URL>\n  --api-key-env <environment variable containing the bearer>\n  --model <provider model id>\n  --limit <positive case count>\n\nControls:\n  --resume                         Continue an interrupted run in --out\n  --seed <integer>                 Case-selection and comparison seed. Default: 0\n  --concurrency <positive integer> Parallel benchmark jobs. Default: 1\n  --repetitions <positive integer> Runs per case and runner. Default: 1\n  --max-output-tokens <positive>   Model output limit per call. Default: 16384\n  --python <executable>             Python with agent-eval-rpc[dspy]. Default: python\n  --timeout-ms <positive>          Model analyst deadline per case. Default: 300000\n  --max-cost-usd <positive>        Run-wide spend limit. Default: 5\n  --max-artifact-bytes <positive>  Final evidence bytes per case. Default: 8388608\n\nWrites result.json with every observation, metric, usage field, error, comparison,\ninput digest, artifact digest, case distribution, selected case id, and explicit\nunknown cost. Limited deterministic-hash subsets are marked non-representative.\nCompleted observations are fsynced to observations.jsonl. Shareable output is in\nresult.json and report.md. Machine-local paths, endpoint, and command are isolated\nin run.local.json.\nThe key is read from the named environment variable and is never written.";
 //#endregion
 //#region src/analyst/benchmark-implementation.d.ts
 declare const ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM = "sha256-canonical-source-manifest";
 declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM = "sha256-canonical-file-manifest";
 declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES: readonly string[];
-declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "11ea62baf2914dba71dd62e45ba4bb8e856b2cefdd5b1b73dc2f14b4b6e6d115";
+declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "1e4c763f2b791c57b37c0e326c56546b433a7a99c0bc7d064abdb58b9cbf0d81";
 /** The published benchmark evidence was produced at this package version, by
  * the retired one-shot direct runner, before trace analysts moved to the
  * recursive DSPy RLM engine. Both evidence digests below are historical facts
@@ -703,7 +705,7 @@ declare const ANALYST_BENCHMARK_EVIDENCE_PACKAGE_VERSION = "0.137.0";
 declare const ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256 = "1e03f2daed356d60316aabefb407ec1e437ac94d408d61eea4ae096e9c6fbb5b";
 declare const ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256 = "4dba263b6256a30d56c7fdb2d992d3a953c0035d731f359b704db806f68f75ac";
 declare const ANALYST_BENCHMARK_IMPLEMENTATION_FILES: readonly string[];
-declare const ANALYST_BENCHMARK_IMPLEMENTATION_SHA256 = "26820de559ca069145eb9649283d2efba51a0a7b8d3662d4c378b9ecebfd4777";
+declare const ANALYST_BENCHMARK_IMPLEMENTATION_SHA256 = "56dd4c7ed19fc5855f99ad238464ad95cb1bea155ad58c49dab3eb0f6cbe7d6a";
 declare function analystBenchmarkImplementationDigest(): string;
 declare function analystBenchmarkDependencyLockDigest(): string;
 //#endregion

package/dist/analyst/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","names":[],"sources":["../../src/analyst/adapters.ts","../../src/analyst/benchmark-agentrx-calibration.ts","../../src/analyst/benchmark-dataset-types.ts","../../src/analyst/benchmark-dataset-agentrx.ts","../../src/analyst/benchmark-dataset-codetrace.ts","../../src/analyst/benchmark-dataset-utils.ts","../../src/analyst/benchmark-verification-outcome.ts","../../src/analyst/benchmark-verification-artifacts.ts","../../src/analyst/benchmark-public-types.ts","../../src/analyst/benchmark-public-adapters.ts","../../src/analyst/benchmark-public-data.ts","../../src/analyst/benchmark-public-prompt.ts","../../src/analyst/benchmark-public-model.ts","../../src/analyst/benchmark-public-rlm.ts","../../src/analyst/benchmark-comparison.ts","../../src/analyst/benchmark-public-calibration.ts","../../src/analyst/benchmark-command-artifact.ts","../../src/analyst/benchmark-command-result.ts","../../src/analyst/benchmark-command.ts","../../src/analyst/benchmark-implementation.ts","../../src/analyst/benchmark-report.ts","../../src/analyst/parse-tolerant.ts","../../src/analyst/proposal-findings.ts"],"mappings":";;;;;;;;;;;;iBA4CgB,aAAa,GAAG,WAAgB;UAe/B,oBAAoB;EACnC;EACA;EACA,UAAU,mBAAmB;;;;;EAK7B,UAAU,KAAK,cAAc;;iBAGf,sBAAsB,KAAK,MAAM,oBAAoB,OAAO,QAAQ;UAwEnE;EACf;EACA;EACA,SAAS;;EAET;;iBAGc,uBAAuB,OAAM,uBAA4B,QAAQ;UAiEhE;EACf;EACA;EACA,OAAO;;EAEP,MAAM;;EAEN,OAAO;;EAEP;;iBAGc,mBAAmB,MAAM,mBAAmB,QAAQ;UAiDnD;EACf;EACA;;EAEA,UAAU,KAAK;;EAEf;;iBAGc,kCACd,OAAM,kCACL,QAAQ;;;cC5RE;UAEI;EACf;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA,SAAS;;iBAGK,4BACd,QAAQ,wBACR,2BACC;iBAkBa,iCAAiC,SAAS;;;KCtD9C;UAEK;EACf,YAAY;EACZ;EACA;EACA;EACA;EACA;;UAGe;EACf,eAAe;EACf,mBAAmB;EACnB;IAAe,YAAY;IAAY;;EACvC,wBAAwB;EACxB;EACA;EACA;;UAGe;EACf,UAAU;EACV;EACA;EACA;EACA;;UAGe;EACf,UAAU;EACV,mBAAmB;EACnB;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA,iBAAiB;;KAGP,0CAEC,sCACA,iCACA;UAEI;EACf;EACA;EACA;EACA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA,oCAAoC;;UAGrB;EACf,eAAe;EACf,WAAW,sBAAsB;;KAGvB;UAEK;;;;;EAKf,WAAW;;UAGI,kCACP,yBACN;UAEa,oCAAoC;EACnD;;EAEA;;UAGe,yCAAyC;EACxD;EACA;EACA;EACA;;UAGe,2CACP,kCACN;;;iBCtGY,qBAAqB,QACnC,KAAK,YACL,OAAO,QACP,UAAS,8BACR,qBAAqB;;iBAoGR,6BACd,mBAAmB,YACnB,iBACA,UAAS,mCACR;iBA6Ea,yBAAyB;;iBAoNzB,iBAAiB;;;iBC7YjB,mBAAmB,QACjC,KAAK,mBACL,OAAO,QACP,UAAS,4BACR,qBAAqB;;iBAwFR,gCACd,2BACA,aAAa,uBACb,UAAS,qCACR;;;iBClHa,wBAAwB;;;KCA5B;UAEK;EACf;EACA;EACA,QAAQ;;UAGO;EACf,QAAQ;EACR;EAKA;IAAe;IAAe;;EAC9B,SAAS;EACT;EACA;EACA;EACA;;UAGe;EACf;EACA;;iBA8Ec,yBACd,gBAAgB,2BACf;;;cChGU;KAED;UAEK;EACf,MAAM;EACN;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA,SAAS;EACT;EACA;EACA;EACA;EACA;EACA,OAAO;EACP,cAAc;EACd,UAAU,OAAO;;UAGF;EACf,UAAU;EACV,SAAS;EACT,OAAO,MAAM;IAA6B;;;iBAatB,mCAAmC;EACvD;EACA,KAAK;EACL;IACE,QAAQ;iBAwHI,kCACd,kBACA,iBACA,WAAW,6BACX;;;KChLU;UAEK;EACf;EACA;EACA;EACA;EACA;;EAEA,UAAU;;EAEV;EACA;IACE,SAAS;IACT;IACA;IACA;IACA;;EAEF,aAAa;EACb;IACE;IACA;;;EAGF,mBAAmB;;UAGJ;EACf,OAAO,qBAAqB;EAC5B;EACA;EACA;EACA,YAAY;IACV;IACA;IACA;;EAEF,uBAAuB;EACvB,WAAW;;UAGI;EACf;EACA;EACA,QAAQ;;UAGO;EACf,OAAO;EACP,OAAO;EACP,OAAO;EACP,YAAY;EACZ,QAAQ;;UAGO;EACf;EACA;EACA;EACA;EACA;EACA;EACA,QAAQ;EACR,UAAU;;;;;;;;;;UCnDK;EACf;EACA;;EAEA;EACA;EACA,UAAU;EACV;EACA;EACA;EACA;EACA,WAAW;;;;;;;;;UAUI;EACf;EACA;;EAEA,kCAAkC;;EAElC;;EAEA;;iBAGc,8BAA8B,uBAAuB;iBAiB/C,6BAA6B;EACjD,SAAS;EACT;EACA,mBAAmB;EACnB;EACA,OAAO;EACP,SAAS;IACP;EAAU,UAAU;EAAkB,aAAa;;;;;;;;;iBAqJjC,6BAA6B;EACjD;EACA,iBAAiB;EACjB,OAAO;EACP;EACA;EACA,SAAS;IACP;EAAU,UAAU;EAAkB,aAAa;;;;iBCvLjC,wBACpB,eACC,QAAQ,MAAM;iBA2BD,0BACd,SAAS,+BACT,eAAe,2BACf;EAAW;EAAe;IACzB,MAAM;iBAwBO,6BACd,SAAS,+BACT,eAAe,4BACd;iBAyCa,+BACd,SAAS,+BACT,iBAAiB,2BACjB,mBAAmB,2BACnB,eACC;iBAcmB,8BAA8B;EAClD,SAAS;EACT;EACA;EACA;EACA;EACA;EACA;IACE,QAAQ;;;;;;;cCxKC;;;;cAKA;cAMA;;iBAyFG,4BAA4B,SAAS;;iBAerC,+BAA+B,SAAS;;;;iBAaxC,8BAA8B,SAAS;;;;iBCxEvC,kCACd,SAAS,+BACT,QAAQ,oCACP,uBAAuB;;;;iBCtCV,+BACd,SAAS,+BACT,QAAQ,oCACP,uBAAuB;;;KC7Bd;UAoBK;EACf,QAAQ;EACR;;EAEA;;EAEA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA,SAAS;;iBASK,sBACd,QAAQ,wBACR;EACE;EACA;EACA;EACA;EACA;IAED;;;UCnEc;EACf;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA,SAAS;;iBAGK,8BACd,QAAQ,yBACP;iBAca,mCAAmC,SAAS;;;UCrC3C;EACf;EACA;EACA;IACE,SAAS;IACT;IACA;IACA;IACA;IACA,YAAY;MAAQ;MAAiB;MAAsB;;IAC3D,uBAAuB;IACvB,0BAA0B;IAC1B;MACE;MACA;MACA;MACA,QAAQ;;IAEV;MACE;MACA;MACA;MACA;MACA;MACA;MACA;MACA;MACA;MACA;;;EAGJ,QAAQ;EACR,aAAa;EACb,uBAAuB;EACvB,qBAAqB;;UAGN;EACf;EACA;EACA;EACA;IACE;IACA;IACA;;;UAIa;EACf;IACE,SAAS;IACT;IACA;IACA;MACE;MACA;MACA;;IAEF;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;;EAEF;IACE;IACA;IACA;IACA,YAAY;MAAQ;MAAiB;MAAsB;;IAC3D;IACA;;;UAIa;EACf;EACA;EACA;EACA;EACA,UAAU;;UAGK;EACf;EACA;EACA;EACA;IACE;IACA;IACA;IACA;IACA;IACA;;EAEF;EACA;IACE;IACA;IACA;;EAEF;IACE;IACA;IACA;IACA;IACA;IACA;;;UAIa;EACf;EACA;EACA;EACA,aAAa;EACb;;cAGW;cACA;cACA;cACA;;;iBCxHS,6BACpB,eACC,QAAQ;;;UC0DM;EACf,uBACE,SAAS,+BACT,QAAQ,sCACL,uBAAuB;;;;;;;;;;KAWlB;UAEK;EACf,SAAS;EACT,SAAS;EACT;EACA;EACA;EACA;EACA;EACA;EACA,OAAO;EACP;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;iBAKoB,2BACpB,yBACA,MAAK,OAAO,YACZ,eAAc,sCACb;cAmRU;;;cC9YA;cAEA;cAEA;cAOA;;;;;;;;cAUA;cAEA;cAGA;cAGA;cAoFA;iBAGG;iBAIA;;;iBCrHA,+BACd,QAAQ,wBACR,uBAAsB;;;;;;;;;;;;;;;iBCQR,gBAAgB;;;;;iBAgBhB,WAAW;;;;;;;iBAeX,oBAAoB;;;;iBCXpB,kBAAkB,mBAAmB,WAAW;;;;;;iBAShD,uBACd,mBACA,mBACC,cAAc"}
+{"version":3,"file":"index.d.ts","names":[],"sources":["../../src/analyst/adapters.ts","../../src/analyst/benchmark-agentrx-calibration.ts","../../src/analyst/benchmark-dataset-types.ts","../../src/analyst/benchmark-dataset-agentrx.ts","../../src/analyst/benchmark-dataset-codetrace.ts","../../src/analyst/benchmark-dataset-utils.ts","../../src/analyst/benchmark-verification-outcome.ts","../../src/analyst/benchmark-verification-artifacts.ts","../../src/analyst/benchmark-public-types.ts","../../src/analyst/benchmark-public-adapters.ts","../../src/analyst/benchmark-public-data.ts","../../src/analyst/benchmark-public-prompt.ts","../../src/analyst/benchmark-public-model.ts","../../src/analyst/benchmark-public-rlm.ts","../../src/analyst/benchmark-comparison.ts","../../src/analyst/benchmark-public-calibration.ts","../../src/analyst/benchmark-command-artifact.ts","../../src/analyst/benchmark-command-result.ts","../../src/analyst/benchmark-command.ts","../../src/analyst/benchmark-implementation.ts","../../src/analyst/benchmark-report.ts","../../src/analyst/parse-tolerant.ts","../../src/analyst/proposal-findings.ts"],"mappings":";;;;;;;;;;;;iBA4CgB,aAAa,GAAG,WAAgB;UAe/B,oBAAoB;EACnC;EACA;EACA,UAAU,mBAAmB;;;;;EAK7B,UAAU,KAAK,cAAc;;iBAGf,sBAAsB,KAAK,MAAM,oBAAoB,OAAO,QAAQ;UAwEnE;EACf;EACA;EACA,SAAS;;EAET;;iBAGc,uBAAuB,OAAM,uBAA4B,QAAQ;UAiEhE;EACf;EACA;EACA,OAAO;;EAEP,MAAM;;EAEN,OAAO;;EAEP;;iBAGc,mBAAmB,MAAM,mBAAmB,QAAQ;UAiDnD;EACf;EACA;;EAEA,UAAU,KAAK;;EAEf;;iBAGc,kCACd,OAAM,kCACL,QAAQ;;;cC5RE;UAEI;EACf;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA,SAAS;;iBAGK,4BACd,QAAQ,wBACR,2BACC;iBAkBa,iCAAiC,SAAS;;;KCtD9C;UAEK;EACf,YAAY;EACZ;EACA;EACA;EACA;EACA;;UAGe;EACf,eAAe;EACf,mBAAmB;EACnB;IAAe,YAAY;IAAY;;EACvC,wBAAwB;EACxB;EACA;EACA;;UAGe;EACf,UAAU;EACV;EACA;EACA;EACA;;UAGe;EACf,UAAU;EACV,mBAAmB;EACnB;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA,iBAAiB;;KAGP,0CAEC,sCACA,iCACA;UAEI;EACf;EACA;EACA;EACA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA,oCAAoC;;UAGrB;EACf,eAAe;EACf,WAAW,sBAAsB;;KAGvB;UAEK;;;;;EAKf,WAAW;;UAGI,kCACP,yBACN;UAEa,oCAAoC;EACnD;;EAEA;;UAGe,yCAAyC;EACxD;EACA;EACA;EACA;;UAGe,2CACP,kCACN;;;iBCtGY,qBAAqB,QACnC,KAAK,YACL,OAAO,QACP,UAAS,8BACR,qBAAqB;;iBAoGR,6BACd,mBAAmB,YACnB,iBACA,UAAS,mCACR;iBA6Ea,yBAAyB;;iBAoNzB,iBAAiB;;;iBC7YjB,mBAAmB,QACjC,KAAK,mBACL,OAAO,QACP,UAAS,4BACR,qBAAqB;;iBAwFR,gCACd,2BACA,aAAa,uBACb,UAAS,qCACR;;;iBClHa,wBAAwB;;;KCA5B;UAEK;EACf;EACA;EACA,QAAQ;;UAGO;EACf,QAAQ;EACR;EAKA;IAAe;IAAe;;EAC9B,SAAS;EACT;EACA;EACA;EACA;;UAGe;EACf;EACA;;iBA8Ec,yBACd,gBAAgB,2BACf;;;cChGU;KAED;UAEK;EACf,MAAM;EACN;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA,SAAS;EACT;EACA;EACA;EACA;EACA;EACA,OAAO;EACP,cAAc;EACd,UAAU,OAAO;;UAGF;EACf,UAAU;EACV,SAAS;EACT,OAAO,MAAM;IAA6B;;;iBAatB,mCAAmC;EACvD;EACA,KAAK;EACL;IACE,QAAQ;iBAwHI,kCACd,kBACA,iBACA,WAAW,6BACX;;;KChLU;UAEK;EACf;EACA;EACA;EACA;EACA;;EAEA,UAAU;;EAEV;EACA;IACE,SAAS;IACT;IACA;IACA;IACA;;EAEF,aAAa;EACb;IACE;IACA;;;EAGF,mBAAmB;;UAGJ;EACf,OAAO,qBAAqB;EAC5B;EACA;EACA;EACA,YAAY;IACV;IACA;IACA;;EAEF,uBAAuB;EACvB,WAAW;;UAGI;EACf;EACA;EACA,QAAQ;;UAGO;EACf,OAAO;EACP,OAAO;EACP,OAAO;EACP,YAAY;EACZ,QAAQ;;UAGO;EACf;EACA;EACA;EACA;EACA;EACA;EACA,QAAQ;EACR,UAAU;;;;;;;;;;UCnDK;EACf;EACA;;EAEA;EACA;EACA,UAAU;EACV;EACA;EACA;EACA;EACA,WAAW;;;;;;;;;UAUI;EACf;EACA;;EAEA,kCAAkC;;EAElC;;EAEA;;EAEA;;iBAGc,8BAA8B,uBAAuB;iBAiB/C,6BAA6B;EACjD,SAAS;EACT;EACA,mBAAmB;EACnB;EACA,OAAO;EACP,SAAS;IACP;EAAU,UAAU;EAAkB,aAAa;;;;;;;;;iBAqKjC,6BAA6B;EACjD;EACA,iBAAiB;EACjB,OAAO;EACP;EACA;EACA,SAAS;IACP;EAAU,UAAU;EAAkB,aAAa;;;;iBCzMjC,wBACpB,eACC,QAAQ,MAAM;iBA2BD,0BACd,SAAS,+BACT,eAAe,2BACf;EAAW;EAAe;IACzB,MAAM;iBAwBO,6BACd,SAAS,+BACT,eAAe,4BACd;iBAyCa,+BACd,SAAS,+BACT,iBAAiB,2BACjB,mBAAmB,2BACnB,eACC;iBAcmB,8BAA8B;EAClD,SAAS;EACT;EACA;EACA;EACA;EACA;EACA;IACE,QAAQ;;;;;;;cCxKC;;;;cAKA;cAMA;;iBA0FG,4BAA4B,SAAS;;iBAerC,+BAA+B,SAAS;;;;iBAaxC,8BAA8B,SAAS;;;;iBCzEvC,kCACd,SAAS,+BACT,QAAQ,oCACP,uBAAuB;;;;iBCtCV,+BACd,SAAS,+BACT,QAAQ,oCACP,uBAAuB;;;KC7Bd;UAoBK;EACf,QAAQ;EACR;;EAEA;;EAEA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA,SAAS;;iBASK,sBACd,QAAQ,wBACR;EACE;EACA;EACA;EACA;EACA;IAED;;;UCnEc;EACf;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA,SAAS;;iBAGK,8BACd,QAAQ,yBACP;iBAca,mCAAmC,SAAS;;;UCrC3C;EACf;EACA;EACA;IACE,SAAS;IACT;IACA;IACA;IACA;IACA,YAAY;MAAQ;MAAiB;MAAsB;;IAC3D,uBAAuB;IACvB,0BAA0B;IAC1B;MACE;MACA;MACA;MACA,QAAQ;;IAEV;MACE;MACA;MACA;MACA;MACA;MACA;MACA;MACA;MACA;MACA;;;EAGJ,QAAQ;EACR,aAAa;EACb,uBAAuB;EACvB,qBAAqB;;UAGN;EACf;EACA;EACA;EACA;IACE;IACA;IACA;;;UAIa;EACf;IACE,SAAS;IACT;IACA;IACA;MACE;MACA;MACA;;IAEF;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;;EAEF;IACE;IACA;IACA;IACA,YAAY;MAAQ;MAAiB;MAAsB;;IAC3D;IACA;;;UAIa;EACf;EACA;EACA;EACA;EACA,UAAU;;UAGK;EACf;EACA;EACA;EACA;IACE;IACA;IACA;IACA;IACA;IACA;;EAEF;EACA;IACE;IACA;IACA;;EAEF;IACE;IACA;IACA;IACA;IACA;IACA;;;UAIa;EACf;EACA;EACA;EACA,aAAa;EACb;;cAGW;cACA;cACA;cACA;;;iBCxHS,6BACpB,eACC,QAAQ;;;UC8DM;EACf,uBACE,SAAS,+BACT,QAAQ,sCACL,uBAAuB;;;;;;;;;;KAWlB;UAEK;EACf,SAAS;EACT,SAAS;EACT;EACA;EACA;EACA;EACA;EACA;EACA,OAAO;EACP;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;iBAKoB,2BACpB,yBACA,MAAK,OAAO,YACZ,eAAc,sCACb;cAsSU;;;cCraA;cAEA;cAEA;cAOA;;;;;;;;cAUA;cAEA;cAGA;cAGA;cAoFA;iBAGG;iBAIA;;;iBCrHA,+BACd,QAAQ,wBACR,uBAAsB;;;;;;;;;;;;;;;iBCQR,gBAAgB;;;;;iBAgBhB,WAAW;;;;;;;iBAeX,oBAAoB;;;;iBCXpB,kBAAkB,mBAAmB,WAAW;;;;;;iBAShD,uBACd,mBACA,mBACC,cAAc"}

package/dist/analyst/index.js

@@ -1,12 +1,12 @@
-import { i as CostLedger } from "../cost-ledger-CZ9diLxY.js";
-import { C as createChatClient, _ as CONTROL_INTEGRITY_ANALYST, b as behavioralAnalyst, f as DEFAULT_TRACE_ANALYST_KINDS, g as FAILURE_MODE_KIND_SPEC, h as IMPROVEMENT_KIND_SPEC, i as assertExactRegistryRunOpts, m as KNOWLEDGE_GAP_KIND_SPEC, n as AnalystRegistry, p as KNOWLEDGE_POISONING_KIND_SPEC, r as ExactAnalystRunExecutionError, t as buildDefaultAnalystRegistry, v as ControlIntegrityAnalyst, x as deriveEfficiencyFindings, y as emitControlIntegrityFindings } from "../default-registry-BgJJItGr.js";
-import { A as coerceJson, B as renderFindingSubject, D as RawAnalystFindingSchema, E as RawAnalystEvidenceSchema, F as FINDING_SUBJECT_SYNTAX, G as resolveTraceAnalystLimits, I as FindingSubjectStringSchema, L as KIND_EXPECTED_SUBJECTS, M as stripCodeFences, N as FINDING_SUBJECT_GRAMMAR_PROMPT, O as evidenceRefsFromRawFinding, P as FINDING_SUBJECT_KINDS, R as findingSubjectGrammarPromptFor, T as RAW_FINDING_SCHEMA_PROMPT, W as DEFAULT_TRACE_ANALYST_LIMITS, a as buildTraceToolsForGroup, i as runTraceAnalyst, j as coerceToFindingRows, k as parseRawFinding, n as renderPriorFindings, r as renderUpstreamFindings, t as createTraceAnalyst, w as ANALYST_SEVERITIES, z as parseFindingSubject } from "../kind-factory-CFxA0JQX.js";
+import { i as CostLedger } from "../cost-ledger-DMFxsLKr.js";
+import { C as createChatClient, _ as CONTROL_INTEGRITY_ANALYST, b as behavioralAnalyst, f as DEFAULT_TRACE_ANALYST_KINDS, g as FAILURE_MODE_KIND_SPEC, h as IMPROVEMENT_KIND_SPEC, i as assertExactRegistryRunOpts, m as KNOWLEDGE_GAP_KIND_SPEC, n as AnalystRegistry, p as KNOWLEDGE_POISONING_KIND_SPEC, r as ExactAnalystRunExecutionError, t as buildDefaultAnalystRegistry, v as ControlIntegrityAnalyst, x as deriveEfficiencyFindings, y as emitControlIntegrityFindings } from "../default-registry-DCp-6hc-.js";
+import { A as coerceJson, B as renderFindingSubject, D as RawAnalystFindingSchema, E as RawAnalystEvidenceSchema, F as FINDING_SUBJECT_SYNTAX, G as resolveTraceAnalystLimits, I as FindingSubjectStringSchema, L as KIND_EXPECTED_SUBJECTS, M as stripCodeFences, N as FINDING_SUBJECT_GRAMMAR_PROMPT, O as evidenceRefsFromRawFinding, P as FINDING_SUBJECT_KINDS, R as findingSubjectGrammarPromptFor, T as RAW_FINDING_SCHEMA_PROMPT, W as DEFAULT_TRACE_ANALYST_LIMITS, a as buildTraceToolsForGroup, i as runTraceAnalyst, j as coerceToFindingRows, k as parseRawFinding, n as renderPriorFindings, r as renderUpstreamFindings, t as createTraceAnalyst, w as ANALYST_SEVERITIES, z as parseFindingSubject } from "../kind-factory-DB7nIs35.js";
 import { a as computeFindingId, i as validateUsageSettlementTimeout, n as settleUsageReceiptFromCostLedger, o as makeFinding, s as makeProposalFinding } from "../usage-receipt-CgxMEBZq.js";
 import { n as isProposalFinding, t as assertProposalFindings } from "../proposal-findings-2GIUo1et.js";
-import { a as RunCritic, c as buildSkillUsageReport, d as defaultIsMaterial, f as diffFindings, h as defineTraceAnalyst, i as runSemanticConceptJudge, l as emitSkillUsageFindings, m as defineCustomAnalyst, n as SEMANTIC_CONCEPT_JUDGE_VERSION, o as SKILL_USAGE_ANALYST, s as SkillUsageAnalyst, u as FindingsStore } from "../semantic-concept-judge-BuIJ9IfB.js";
-import { t as createDspyRlmTraceEngine } from "../dspy-rlm-engine-DTkVyDX-.js";
+import { a as RunCritic, c as buildSkillUsageReport, d as defaultIsMaterial, f as diffFindings, h as defineTraceAnalyst, i as runSemanticConceptJudge, l as emitSkillUsageFindings, m as defineCustomAnalyst, n as SEMANTIC_CONCEPT_JUDGE_VERSION, o as SKILL_USAGE_ANALYST, s as SkillUsageAnalyst, u as FindingsStore } from "../semantic-concept-judge-DJQtFr95.js";
+import { t as createDspyRlmTraceEngine } from "../dspy-rlm-engine-Bkak4nzo.js";
 import { a as scoreAnalystFindings, n as runAnalystBenchmark, r as traceStoreEvidenceResolver, t as registryBenchmarkRunner } from "../benchmark-CYtcIF2V.js";
-import { A as ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256, B as ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE, C as publicBenchmarkSystemPrompt, D as parseVerificationOutcome, E as loadCodeTraceVerificationArtifacts, F as ANALYST_BENCHMARK_IMPLEMENTATION_FILES, G as summarizeAgentRxCalibration, H as ANALYST_BENCHMARK_OBSERVATIONS_FILE, I as ANALYST_BENCHMARK_IMPLEMENTATION_SHA256, J as agentRxBenchmarkCase, K as codeTraceBenchCase, L as analystBenchmarkDependencyLockDigest, M as ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256, N as ANALYST_BENCHMARK_EVIDENCE_PACKAGE_VERSION, O as ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM, P as ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM, Q as normalizeBenchmarkLabel, R as analystBenchmarkImplementationDigest, S as publicBenchmarkRlmInstructions, T as appendVerificationArtifactsToOtlp, U as AGENT_RX_UPSTREAM_REVISION, V as ANALYST_BENCHMARK_MANIFEST_FILE, W as renderAgentRxCalibrationMarkdown, X as normalizeAgentRxCategory, Y as agentRxPredictionsToFindings, Z as roundAgentRxStep, _ as expandCodeTraceFailureBlocks, a as renderCodeTraceCalibrationMarkdown, b as MAX_INCORRECT_BLOCK_STEPS, c as createPublicBenchmarkRlmRunner, d as preparePublicAnalystBenchmark, f as publicBenchmarkDistributions, g as emptyPublicBenchmarkRunner, h as adaptPublicBenchmarkFindings, i as readAnalystBenchmarkArtifact, j as ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256, k as ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES, l as createPublicBenchmarkDirectRunner, m as selectPublicBenchmarkRows, n as runAnalystBenchmarkCommand, o as summarizeCodeTraceCalibration, p as publicBenchmarkSelectionReport, q as codeTracerPredictionsToFindings, r as renderAnalystBenchmarkMarkdown, s as compareAnalystRunners, t as ANALYST_BENCHMARK_HELP, u as loadPublicBenchmarkRows, v as CODE_TRACE_BENCH_ANALYST_PROMPT, w as DEFAULT_MAX_VERIFICATION_ARTIFACT_BYTES, x as publicBenchmarkProtocolSha256, y as MAX_INCORRECT_BLOCKS, z as ANALYST_BENCHMARK_COST_LEDGER_FILE } from "../benchmark-command-xvi2liH7.js";
+import { A as ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256, B as ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE, C as publicBenchmarkSystemPrompt, D as parseVerificationOutcome, E as loadCodeTraceVerificationArtifacts, F as ANALYST_BENCHMARK_IMPLEMENTATION_FILES, G as summarizeAgentRxCalibration, H as ANALYST_BENCHMARK_OBSERVATIONS_FILE, I as ANALYST_BENCHMARK_IMPLEMENTATION_SHA256, J as agentRxBenchmarkCase, K as codeTraceBenchCase, L as analystBenchmarkDependencyLockDigest, M as ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256, N as ANALYST_BENCHMARK_EVIDENCE_PACKAGE_VERSION, O as ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM, P as ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM, Q as normalizeBenchmarkLabel, R as analystBenchmarkImplementationDigest, S as publicBenchmarkRlmInstructions, T as appendVerificationArtifactsToOtlp, U as AGENT_RX_UPSTREAM_REVISION, V as ANALYST_BENCHMARK_MANIFEST_FILE, W as renderAgentRxCalibrationMarkdown, X as normalizeAgentRxCategory, Y as agentRxPredictionsToFindings, Z as roundAgentRxStep, _ as expandCodeTraceFailureBlocks, a as renderCodeTraceCalibrationMarkdown, b as MAX_INCORRECT_BLOCK_STEPS, c as createPublicBenchmarkRlmRunner, d as preparePublicAnalystBenchmark, f as publicBenchmarkDistributions, g as emptyPublicBenchmarkRunner, h as adaptPublicBenchmarkFindings, i as readAnalystBenchmarkArtifact, j as ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256, k as ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES, l as createPublicBenchmarkDirectRunner, m as selectPublicBenchmarkRows, n as runAnalystBenchmarkCommand, o as summarizeCodeTraceCalibration, p as publicBenchmarkSelectionReport, q as codeTracerPredictionsToFindings, r as renderAnalystBenchmarkMarkdown, s as compareAnalystRunners, t as ANALYST_BENCHMARK_HELP, u as loadPublicBenchmarkRows, v as CODE_TRACE_BENCH_ANALYST_PROMPT, w as DEFAULT_MAX_VERIFICATION_ARTIFACT_BYTES, x as publicBenchmarkProtocolSha256, y as MAX_INCORRECT_BLOCKS, z as ANALYST_BENCHMARK_COST_LEDGER_FILE } from "../benchmark-command-CK0UnXAD.js";
 //#region src/analyst/adapters.ts
 /**
 * Adapter factories — lift each existing agent-eval primitive into the

package/dist/{benchmark-DDVdWcwA.d.ts → benchmark-DxaZfy0w.d.ts}

@@ -1,5 +1,5 @@
-import { c as AnalystRunInputs, f as AnalystUsageReceipt, i as AnalystFinding, p as EvidenceRef, x as TraceAnalysisStore } from "./types-DoEYskCd.js";
-import { E as AnalystRegistry, M as RegistryRunOpts } from "./completion-verifier-IPoP4fQO.js";
+import { c as AnalystRunInputs, f as AnalystUsageReceipt, i as AnalystFinding, p as EvidenceRef, x as TraceAnalysisStore } from "./types-D3jh6F98.js";
+import { E as AnalystRegistry, M as RegistryRunOpts } from "./completion-verifier-D15NHYSk.js";
 //#region src/analyst/benchmark-scoring.d.ts
 declare function scoreAnalystFindings(testCase: Pick<AnalystBenchmarkCase, 'id' | 'expectedIssues' | 'labeledEvidence'>, findings: readonly AnalystFinding[]): AnalystFindingScore;
 //#endregion
@@ -233,4 +233,4 @@ declare function registryBenchmarkRunner(options: {
 }): AnalystBenchmarkRunner<AnalystRunInputs>;
 //#endregion
 export { scoreAnalystFindings as C, traceStoreEvidenceResolver as S, AnalystIssueExpectation as _, AnalystBenchmarkLabelState as a, registryBenchmarkRunner as b, AnalystBenchmarkProvenance as c, AnalystBenchmarkSummary as d, AnalystEvidenceExpectation as f, AnalystFindingScore as g, AnalystEvidenceResolver as h, AnalystBenchmarkError as i, AnalystBenchmarkResult as l, AnalystEvidenceResolutionError as m, AnalystBenchmarkDatasetRef as n, AnalystBenchmarkObservation as o, AnalystEvidenceResolution as p, AnalystBenchmarkDescriptor as r, AnalystBenchmarkOutput as s, AnalystBenchmarkCase as t, AnalystBenchmarkRunner as u, AnalystLatencyDistribution as v, runAnalystBenchmark as x, RunAnalystBenchmarkOptions as y };
-//# sourceMappingURL=benchmark-DDVdWcwA.d.ts.map
+//# sourceMappingURL=benchmark-DxaZfy0w.d.ts.map

package/dist/{benchmark-DDVdWcwA.d.ts.map → benchmark-DxaZfy0w.d.ts.map}

@@ -1 +1 @@
-{"version":3,"file":"benchmark-DDVdWcwA.d.ts","names":[],"sources":["../src/analyst/benchmark-scoring.ts","../src/analyst/benchmark.ts"],"mappings":";;;iBASgB,qBACd,UAAU,KAAK,oEACf,mBAAmB,mBAClB;;;UCIc;EACf;EACA,OAAO;;UAGQ;EACf;EACA;EACA;EACA;EACA,oBAAoB;EACpB;;EAEA,4BAA4B;;KAGlB;UAEK,qBAAqB;EACpC;;EAEA;;EAEA,YAAY;EACZ,OAAO;EACP,yBAAyB;;EAEzB,2BAA2B;EAC3B;EACA,WAAW;;UAGI;EACf;EACA;EACA;EACA;EACA;EACA,mBAAmB;EACnB;EACA;EACA;EACA;;EAEA;;EAEA;;EAEA;EACA;;UAGe;EACf,UAAU;EACV;EACA;;UAGe;EACf;EACA;EACA,oBAAoB;EACpB,QAAQ;;EAER;;KAGU,wBAAwB,qBAAqB;EACvD;EACA,WAAW;EACX,UAAU;EACV,SAAS;gBACK;;;;;iBAMA,2BAA2B,QACzC,WAAW,OAAO,WAAW,qBAC5B,wBAAwB;UAoBV;EACf,mBAAmB;EACnB,QAAQ;EACR,WAAW;;;;;EAKX;;EAEA,QAAQ;;UAGO;EACf;EACA;EACA;EACA;;UAGe,uBAAuB;EACtC;EACA,QACE,OAAO,QACP;IAAW;IAAgB;IAAoB,SAAS;MACvD,yBAAyB,QAAQ;;UAGrB;EACf;EACA;EACA;EACA,YAAY;EACZ;EACA;EACA;EACA;EACA,mBAAmB;EACnB,OAAO;EACP,qBAAqB;EACrB;EACA,eAAe;EACf,QAAQ;EACR,iBAAiB;EACjB,QAAQ;;UAGO;EACf;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA;EACA;EACA;EACA;;EAEA;;EAEA;;EAEA;;EAEA;;EAEA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;EAEA;;EAEA;;EAEA;;EAEA;EACA,WAAW;EACX;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;;UAGe;EACf;EACA,UAAU;EACV;EACA,cAAc;EACd,WAAW;;UAGI,mCAAmC;EAClD;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf,YAAY;EACZ,cAAc;EACd,WAAW;;UAGI,2BAA2B;EAC1C,gBAAgB,qBAAqB;EACrC,kBAAkB,uBAAuB;EACzC;EACA;EACA;EACA,kBAAkB,wBAAwB;EAC1C,YAAY;;EAEZ,+BAA+B;EAC/B,iBAAiB,aAAa,uCAAuC;EACrE,SAAS;;iBAGW,oBAAoB,QACxC,SAAS,2BAA2B,UACnC,QAAQ;iBAyDK,wBAAwB;EACtC;EACA,UAAU;EACV,aAAa,KAAK;;EAElB;IACE,uBAAuB"}
+{"version":3,"file":"benchmark-DxaZfy0w.d.ts","names":[],"sources":["../src/analyst/benchmark-scoring.ts","../src/analyst/benchmark.ts"],"mappings":";;;iBASgB,qBACd,UAAU,KAAK,oEACf,mBAAmB,mBAClB;;;UCIc;EACf;EACA,OAAO;;UAGQ;EACf;EACA;EACA;EACA;EACA,oBAAoB;EACpB;;EAEA,4BAA4B;;KAGlB;UAEK,qBAAqB;EACpC;;EAEA;;EAEA,YAAY;EACZ,OAAO;EACP,yBAAyB;;EAEzB,2BAA2B;EAC3B;EACA,WAAW;;UAGI;EACf;EACA;EACA;EACA;EACA;EACA,mBAAmB;EACnB;EACA;EACA;EACA;;EAEA;;EAEA;;EAEA;EACA;;UAGe;EACf,UAAU;EACV;EACA;;UAGe;EACf;EACA;EACA,oBAAoB;EACpB,QAAQ;;EAER;;KAGU,wBAAwB,qBAAqB;EACvD;EACA,WAAW;EACX,UAAU;EACV,SAAS;gBACK;;;;;iBAMA,2BAA2B,QACzC,WAAW,OAAO,WAAW,qBAC5B,wBAAwB;UAoBV;EACf,mBAAmB;EACnB,QAAQ;EACR,WAAW;;;;;EAKX;;EAEA,QAAQ;;UAGO;EACf;EACA;EACA;EACA;;UAGe,uBAAuB;EACtC;EACA,QACE,OAAO,QACP;IAAW;IAAgB;IAAoB,SAAS;MACvD,yBAAyB,QAAQ;;UAGrB;EACf;EACA;EACA;EACA,YAAY;EACZ;EACA;EACA;EACA;EACA,mBAAmB;EACnB,OAAO;EACP,qBAAqB;EACrB;EACA,eAAe;EACf,QAAQ;EACR,iBAAiB;EACjB,QAAQ;;UAGO;EACf;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA;EACA;EACA;EACA;;EAEA;;EAEA;;EAEA;;EAEA;;EAEA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;EAEA;;EAEA;;EAEA;;EAEA;EACA,WAAW;EACX;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;;UAGe;EACf;EACA,UAAU;EACV;EACA,cAAc;EACd,WAAW;;UAGI,mCAAmC;EAClD;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf,YAAY;EACZ,cAAc;EACd,WAAW;;UAGI,2BAA2B;EAC1C,gBAAgB,qBAAqB;EACrC,kBAAkB,uBAAuB;EACzC;EACA;EACA;EACA,kBAAkB,wBAAwB;EAC1C,YAAY;;EAEZ,+BAA+B;EAC/B,iBAAiB,aAAa,uCAAuC;EACrE,SAAS;;iBAGW,oBAAoB,QACxC,SAAS,2BAA2B,UACnC,QAAQ;iBAyDK,wBAAwB;EACtC;EACA,UAAU;EACV,aAAa,KAAK;;EAElB;IACE,uBAAuB"}

package/dist/{benchmark-command-xvi2liH7.js → benchmark-command-CK0UnXAD.js}

@@ -1,15 +1,15 @@
 import { c as ValidationError, t as AgentEvalError } from "./errors-D-LKuDhb.js";
 import { s as resolveModelPricing } from "./metrics-C9YY1OcL.js";
-import { a as CostLedgerPersistenceError, i as CostLedger, n as CostCallConflictError, o as CostReceiptCaptureError, r as CostCeilingReachedError, s as CostReservationExceededError, t as CostAccountingIncompleteError } from "./cost-ledger-CZ9diLxY.js";
-import { c as callLlmJson, f as maximumChargeForLlmRequest, l as costReceiptFromLlm, r as LlmResponseError, t as LlmCallError, u as costReceiptFromLlmError } from "./llm-client-bkztEfIx.js";
-import { O as evidenceRefsFromRawFinding, h as TRACE_ANALYSIS_LIMITS, i as runTraceAnalyst } from "./kind-factory-CFxA0JQX.js";
+import { a as CostLedgerPersistenceError, i as CostLedger, n as CostCallConflictError, o as CostReceiptCaptureError, r as CostCeilingReachedError, s as CostReservationExceededError, t as CostAccountingIncompleteError } from "./cost-ledger-DMFxsLKr.js";
+import { c as callLlmJson, f as maximumChargeForLlmRequest, l as costReceiptFromLlm, r as LlmResponseError, t as LlmCallError, u as costReceiptFromLlmError } from "./llm-client-B3WXSH5Y.js";
+import { O as evidenceRefsFromRawFinding, h as TRACE_ANALYSIS_LIMITS, i as runTraceAnalyst } from "./kind-factory-DB7nIs35.js";
 import { o as makeFinding, r as usageReceiptFromCostLedger } from "./usage-receipt-CgxMEBZq.js";
 import { i as hashCanonical, r as canonicalString } from "./canonical-D011XM8r.js";
-import { n as createRunCostLedger, r as fsCampaignStorage, t as acquireSingleRunLock } from "./single-run-lock-BTTtPZ9N.js";
-import { t as createDspyRlmTraceEngine } from "./dspy-rlm-engine-DTkVyDX-.js";
+import { n as createRunCostLedger, r as fsCampaignStorage, t as acquireSingleRunLock } from "./single-run-lock-CiQThJxB.js";
+import { t as createDspyRlmTraceEngine } from "./dspy-rlm-engine-Bkak4nzo.js";
 import { c as writeLedgerFileAtomically, s as withLedgerFileLock } from "./ledger-core-Dxz0Rkwa.js";
 import { E as pairedBootstrap } from "./statistics-ByxzSiOM.js";
-import { i as otlpTextToTraceAnalysisStore, r as createOtlpBufferTraceStore, t as DEFAULT_MAX_TRACE_FILE_BYTES } from "./store-otlp-DX4fGIcf.js";
+import { i as otlpTextToTraceAnalysisStore, r as createOtlpBufferTraceStore, t as DEFAULT_MAX_TRACE_FILE_BYTES } from "./store-otlp-vRByAR6h.js";
 import { i as summarizeAnalystBenchmarkRunner, n as runAnalystBenchmark, r as traceStoreEvidenceResolver } from "./benchmark-CYtcIF2V.js";
 import { z } from "zod";
 import { constants, existsSync, lstatSync, readFileSync } from "node:fs";
@@ -1172,7 +1172,7 @@ const ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES = Object.freeze([
 	"package.json",
 	"pnpm-lock.yaml"
 ]);
-const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "11ea62baf2914dba71dd62e45ba4bb8e856b2cefdd5b1b73dc2f14b4b6e6d115";
+const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "1e4c763f2b791c57b37c0e326c56546b433a7a99c0bc7d064abdb58b9cbf0d81";
 /** The published benchmark evidence was produced at this package version, by
 * the retired one-shot direct runner, before trace analysts moved to the
 * recursive DSPy RLM engine. Both evidence digests below are historical facts
@@ -1266,7 +1266,7 @@ const ANALYST_BENCHMARK_IMPLEMENTATION_FILES = Object.freeze([
 	"src/trace/otlp-attributes.ts",
 	"src/trace/raw-provider-sink.ts"
 ]);
-const ANALYST_BENCHMARK_IMPLEMENTATION_SHA256 = "26820de559ca069145eb9649283d2efba51a0a7b8d3662d4c378b9ecebfd4777";
+const ANALYST_BENCHMARK_IMPLEMENTATION_SHA256 = "56dd4c7ed19fc5855f99ad238464ad95cb1bea155ad58c49dab3eb0f6cbe7d6a";
 function analystBenchmarkImplementationDigest() {
 	return ANALYST_BENCHMARK_IMPLEMENTATION_SHA256;
 }
@@ -2178,6 +2178,7 @@ Never print an entire trace, full source file, or more than 12000 characters in
 Build a compact table of assistant step ids, actions, following observations, and final verification.
 Inspect suspicious steps with viewSpans or searchSpan instead of repeatedly printing the table.
 This runner emits no JSON fields, so the block is encoded in the finding's subject.
+Only findings_json is scored; your prose answer is ignored, so every incorrect block you identify must appear as a finding, never only in the answer.
 Emit exactly one finding per contiguous failure block.
 Set the finding's subject to incorrect-steps-<first_step>-<last_step>-<escape_status>-consequence-<consequence_step>, using the same four values the task defines; for a block covering only step 7 that the agent never escaped and whose damage shows at step 9, the subject is incorrect-steps-7-7-unescaped-consequence-9.
 The runner expands the block to one scored step per member and builds every scored citation itself.
@@ -2300,20 +2301,34 @@ async function adaptCodeTraceFindings(trajectoryId, findings, analystId, store,
 			diagnostics: emptyCodeTraceBlockDiagnostics()
 		};
 	}
-	await validateCodeTraceFindingEvidence({
-		trajectoryId,
-		findings,
-		store,
-		...signal ? { signal } : {}
-	});
-	return expandCodeTraceFailureBlocks({
+	const blocks = [];
+	const rejectedFindings = [];
+	for (const source of findings) try {
+		await validateCodeTraceFindingEvidence({
+			trajectoryId,
+			findings: [source],
+			store,
+			...signal ? { signal } : {}
+		});
+		blocks.push(codeTraceBlockFromFinding(trajectoryId, source));
+	} catch (error) {
+		rejectedFindings.push(`${source.finding_id}: ${error instanceof Error ? error.message : String(error)}`);
+	}
+	const expanded = await expandCodeTraceFailureBlocks({
 		trajectoryId,
-		blocks: findings.map((source) => codeTraceBlockFromFinding(trajectoryId, source)),
+		blocks,
 		store,
 		analystId,
 		...findings[0] ? { producedAt: findings[0].produced_at } : {},
 		...signal ? { signal } : {}
 	});
+	return {
+		findings: expanded.findings,
+		diagnostics: {
+			...expanded.diagnostics,
+			rejectedFindings
+		}
+	};
 }
 function codeTraceBlockFromFinding(trajectoryId, source) {
 	const parsed = CODE_TRACE_BLOCK_SUBJECT.exec(source.subject ?? "");
@@ -3406,9 +3421,9 @@ function trajectoryIdFromCaseId$1(dataset, caseId) {
 function createPublicBenchmarkRlmRunner(dataset, config) {
 	const costLedger = config.costLedger ?? new CostLedger();
 	const limits = {
-		maxIterations: config.dspyRlm?.maxIterations ?? 8,
-		maxLlmCalls: config.dspyRlm?.maxLlmCalls ?? 4,
-		maxToolCalls: config.dspyRlm?.maxToolCalls ?? 32,
+		maxIterations: config.dspyRlm?.maxIterations ?? 14,
+		maxLlmCalls: config.dspyRlm?.maxLlmCalls ?? 8,
+		maxToolCalls: config.dspyRlm?.maxToolCalls ?? 80,
 		maxOutputChars: config.dspyRlm?.maxOutputChars ?? 8e3
 	};
 	const pricing = config.pricing ?? pricingForModel(config.model);
@@ -4539,24 +4554,30 @@ const NON_SCORABLE_COST_ERRORS = /* @__PURE__ */ new Set([
 function assertObservationAccountingComplete(observation, costLedger, analystRunnerId) {
 	if (observation.error && NON_SCORABLE_COST_ERRORS.has(observation.error.class)) throw new CostAccountingIncompleteError(`Analyst benchmark stopped before scoring: ${observation.error.message}`);
 	if (observation.runnerId !== analystRunnerId) return;
-	const summary = costLedger.summary({
+	const filter = {
 		channel: "analyst",
 		tags: {
 			benchmarkCaseId: observation.caseId,
 			benchmarkRepetition: String(observation.repetition)
 		}
-	});
-	if (!summary.accountingComplete || summary.pendingCalls > 0 || summary.unresolvedCalls > 0) throw accountingError(costLedger, "the recursive analyst has incomplete cost accounting", {
-		channel: "analyst",
-		tags: {
-			benchmarkCaseId: observation.caseId,
-			benchmarkRepetition: String(observation.repetition)
-		}
-	});
+	};
+	if (!costAccountingIsTrustworthy(costLedger.summary(filter))) throw accountingError(costLedger, "the recursive analyst has incomplete cost accounting", filter);
+}
+const BUDGET_BREACH_REASON = /exceeding its enforced maximum/;
+/**
+* Cost accounting is trustworthy when every call resolved and none breached its
+* budget. A recursive analyst on a real provider will occasionally receive a
+* settled response whose usage the provider omitted; that call is honestly
+* recorded as unknown and excluded from the reported cost, so it does not
+* invalidate a completed run. A call left pending, one lost, or one charged
+* beyond its maximum is a genuine integrity failure and still halts.
+*/
+function costAccountingIsTrustworthy(summary) {
+	if (summary.pendingCalls > 0 || summary.unresolvedCalls > 0) return false;
+	return !summary.incompleteReasons.some((reason) => BUDGET_BREACH_REASON.test(reason));
 }
 function assertCostLedgerFinalizable(costLedger) {
-	const summary = costLedger.summary();
-	if (!summary.accountingComplete || summary.pendingCalls > 0 || summary.unresolvedCalls > 0) throw accountingError(costLedger, "the run has pending or incomplete cost entries");
+	if (!costAccountingIsTrustworthy(costLedger.summary())) throw accountingError(costLedger, "the run has pending or budget-breaching cost entries");
 }
 function accountingError(costLedger, reason, filter) {
 	const details = costLedger.summary(filter).incompleteReasons.slice(0, 3).join("; ");
@@ -4587,7 +4608,7 @@ Controls:
   --seed <integer>                 Case-selection and comparison seed. Default: 0
   --concurrency <positive integer> Parallel benchmark jobs. Default: 1
   --repetitions <positive integer> Runs per case and runner. Default: 1
-  --max-output-tokens <positive>   Model output limit per call. Default: 4096
+  --max-output-tokens <positive>   Model output limit per call. Default: 16384
   --python <executable>             Python with agent-eval-rpc[dspy]. Default: python
   --timeout-ms <positive>          Model analyst deadline per case. Default: 300000
   --max-cost-usd <positive>        Run-wide spend limit. Default: 5
@@ -4628,7 +4649,7 @@ function parseCommandConfig(argv, env) {
 			baseUrl: openAiCompatibleBaseUrl(requiredFlag(flags, "base-url")),
 			apiKey,
 			model: requiredFlag(flags, "model"),
-			maxOutputTokens: positiveFlag(flags, "max-output-tokens", 4096),
+			maxOutputTokens: positiveFlag(flags, "max-output-tokens", 16384),
 			timeoutMs: positiveFlag(flags, "timeout-ms", 3e5),
 			maxCostUsdPerAnalysis: maxCostUsd,
 			...python ? { dspyRlm: { runner: { command: python } } } : {}
@@ -4807,4 +4828,4 @@ function shellQuote(value) {
 //#endregion
 export { ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 as A, ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE as B, publicBenchmarkSystemPrompt as C, parseVerificationOutcome as D, loadCodeTraceVerificationArtifacts as E, ANALYST_BENCHMARK_IMPLEMENTATION_FILES as F, summarizeAgentRxCalibration as G, ANALYST_BENCHMARK_OBSERVATIONS_FILE as H, ANALYST_BENCHMARK_IMPLEMENTATION_SHA256 as I, agentRxBenchmarkCase as J, codeTraceBenchCase as K, analystBenchmarkDependencyLockDigest as L, ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256 as M, ANALYST_BENCHMARK_EVIDENCE_PACKAGE_VERSION as N, ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM as O, ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM as P, normalizeBenchmarkLabel as Q, analystBenchmarkImplementationDigest as R, publicBenchmarkRlmInstructions as S, appendVerificationArtifactsToOtlp as T, AGENT_RX_UPSTREAM_REVISION as U, ANALYST_BENCHMARK_MANIFEST_FILE as V, renderAgentRxCalibrationMarkdown as W, normalizeAgentRxCategory as X, agentRxPredictionsToFindings as Y, roundAgentRxStep as Z, expandCodeTraceFailureBlocks as _, renderCodeTraceCalibrationMarkdown as a, MAX_INCORRECT_BLOCK_STEPS as b, createPublicBenchmarkRlmRunner as c, preparePublicAnalystBenchmark as d, publicBenchmarkDistributions as f, emptyPublicBenchmarkRunner as g, adaptPublicBenchmarkFindings as h, readAnalystBenchmarkArtifact as i, ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256 as j, ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES as k, createPublicBenchmarkDirectRunner as l, selectPublicBenchmarkRows as m, runAnalystBenchmarkCommand as n, summarizeCodeTraceCalibration as o, publicBenchmarkSelectionReport as p, codeTracerPredictionsToFindings as q, renderAnalystBenchmarkMarkdown as r, compareAnalystRunners as s, ANALYST_BENCHMARK_HELP as t, loadPublicBenchmarkRows as u, CODE_TRACE_BENCH_ANALYST_PROMPT as v, DEFAULT_MAX_VERIFICATION_ARTIFACT_BYTES as w, publicBenchmarkProtocolSha256 as x, MAX_INCORRECT_BLOCKS as y, ANALYST_BENCHMARK_COST_LEDGER_FILE as z };
 
-//# sourceMappingURL=benchmark-command-xvi2liH7.js.map
+//# sourceMappingURL=benchmark-command-CK0UnXAD.js.map